Hello,
We've currently encountered a bug in our application related to the isolation of
objects between different transaction running on the same JVM, and we fear the
bug is originating from OJB. A detailed scenario follows:
We are using OJB RC5 and have isolation-level set to serializable.
We are using ObjectCacheOSCacheImpl as the object cache (but we got the same
results with ObjectCacheDefaultImpl).
We launch two threads on the same JVM.
Thread 1 starts Transaction1 (Tx1).
Thread 2 starts Transaction2 (Tx2).
Thread 1 reads Object A (and locks it for read) using Tx1.
Thread 1 locks Object A for write using Tx1.
Thread 1 alters Object A.
Thread 2 reads Object A (and locks it for read) using Tx2.
Thread 1 confirms Tx1.
Thread 2 confirms Tx2.
Is this behaviour expected? According to the lock manganer documentation this
should not be possible. Is anyone having similar problems? Or are we using OJB
incorrectly?
Not sure this is relevante, but we're using the SequenceManagerHighLowImpl
sequence-manager and all our class/table mappings use optimistic locking:
<field-descriptor
name="ackOptLock"
column="ACK_OPT_LOCK"
jdbc-type="INTEGER"
locking="true"
/>
Any help is greatly apriciated.
Thanks in advance,
Luis Cruz
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
